by Steve Levy
Buchheit would eventually leave to focus on start-up investing, but Taylor became Facebook’s chief technical officer. That move signified a difference between the companies. Google liked to give young people massive responsibility, but it also relied on world-class scientists for its operational innovations. It was like a university: the top executives were equivalent to professors. Facebook preferred kids, figuring that what the sharpest undergrads lacked in experience, they would make up for in audacity.
The older company had a massive edge in revenues. Facebook was struggling to develop its own equivalent of AdWords. It would have to be something as organic to the social network as Google’s ad model was to its search product. The bigger threat to Google wouldn’t be measured in dollars, but in the philosophical challenge. Could it be that social networking, rather than algorithmic exploitation of the web’s intelligence, would assume the central role in people’s online lives? Even if that were not the case, Facebook made it clear that every facet of the Internet would benefit from the power of personal connection. Google had been chasing a future forged out of algorithms and science fiction chronicles. Did the key to the future lay in party photos and daily status reports?
The irony was that Google had been present at the explosion of social networking. In a classic case of the company’s corporate ADD, it simply had blown the opportunity to make the most of what it created.
Back in 2002, a young Google engineer named Orkut Buyukkokten had an idea. “My dream was to connect all the Internet users so they can relate to each other,” he later recalled. “It can make such a difference in people’s lives.” Buyukkokten, who had come to Google from his native Turkey via Stanford, decided to use his 20 percent discretionary time developing a cyberspace preserve where the people of the world could intermingle in peace, presumably so their good vibes would go viral. Designed along the lines of the first big social networking site, Friendster—there was no Facebook then—his creation encouraged users to build profiles for themselves. Upon mutual consent, people would bond with each other. Networks of like-minded people would form. Groups of shared interests would emerge. Virtual flowers would bloom. Eventually, wars would end. Buyukkokten coded it up in a weekend and showed it to Marissa Mayer during her weekly office hours. She loved it and assigned one of her APMs to help out.
Buyukkokten wanted to call it Eden, reflecting his vision of a utopian preserve where people could feel safe and trust each other. But eden.com was owned by an opera company that wouldn’t sell its domain name. The owners of paradise.com and utopia.com were similarly unaccommodating. Finally, the product manager and Mayer thought of naming it after its creator. “Orkut.com” belonged to Buyukkokten himself. Google convinced him, and its social networking service was called Orkut.
Was it a sign of the company’s distrust of the insufficiently algorithmic nature of social software that the product was not branded with the Google name? “We wanted to see if it could stand on its own two feet,” says Mayer, a stricture not required from such Google services as Gmail and Google Maps.
In fact, Orkut almost immediately stood tall. Even though the software was available only by invitation—the first users were Googlers who invited their own friends—hundreds of thousands of people signed up in the first month alone. Soon after launch, there was so much activity in the logs that Mayer demanded that the engineers recheck their stats. “We’d never seen anything like it,” says Mayer. “The system just fell over.” Google had to take Orkut down for a couple of days to recover.
During Orkut’s first few months, the global distribution was typical of other products, with half the traffic in the United States and the next biggest chunk, about 8 percent, in Japan. Google’s response was not to pour resources into the product but to observe as Orkut rose or fell on its own. Though there would be exceptions—Android and YouTube, for instance—most Google products, while exactingly rethought and tweaked during the design process, were similarly left to find their way in the world by themselves. Failure was part of Google, one its leaders accepted. The bulk of Google’s efforts on Orkut weren’t focused on design and features that would make the service more useful but rather a rewrite of Orkut’s Windows infrastructure to conform to Google standards so the system would run faster, accommodate growth more smoothly, and resist spam more effectively. (As Orkut became more popular, it came under attack from identity thieves and those who flooded the service with the usual assortment of virility aid ads and Nigerian inheritance announcements.) Meanwhile, the system coughed and sputtered—and impatient users bailed out. At the time, Eric Schmidt chalked up the experience—where Google blew its chance to dominate social networking—as routine collateral damage from the company’s philosophy of quick launches.
“Larry and Sergey’s vision is ‘Let’s get these systems to prove themselves,’” he said in 2004. “After [the engineer] Orkut built this, we said, ‘God knows if this thing’s any good or not.’ So we waited until it crashed from overuse. So then we put another programmer on the project. And now it has many more.” But it was too late—the moment had passed.
Oddly, Orkut became a sensation in Brazil. “In Brazil, Orkut is the Internet and Google is search,” wrote one local journalist, who added that using Orkut was “like putting sugar in your coffee, watching Globo telenovelas, or heading to the beach from Christmas to Carnival.” On a trip to Brazil in 2006, Sergey Brin was asked why, and he responded, “We don’t know—what do you think?” When pressed, Googlers would refer to stereotypes of Carioca sociability, but that didn’t sufficiently explain why Orkut became the social networking choice of this country over other competitors—or why Orkut was so badly left behind in the rest of the world. Marissa Mayer’s personal analysis was based on the Google yardstick of speed. Brazilians, she says, were used to lousy Internet service and thus more tolerant of the delays. “They would just keep sitting there and waiting,” she says.
Orkut was also dominant in India, where it was the number one Google service—ahead of search and Gmail. “There is no second product in India—Orkut is dominant,” said Manu Rekhi, the Orkut India product manager, in 2007. “I’ve seen beggar kids who use their money to get on Orkut.” Mayer also attributed that success to its quick response compared to other services. “Do you know why Orkut took off in India?” she would ask. “Opposite time zone, and no load on the servers at night. Speed matters.” (Why Orkut ruled in Brazil, however, was a mystery never solved.)
In any case, by the time Google switched Orkut’s code base to a speedier infrastructure, Facebook was beginning its rise in the United States. Google never made a serious attempt to dislodge it. In 2008, Google announced that it was moving all of Orkut’s operations to its offices in Belo Horizonte, Brazil. By then about half of Orkut’s traffic came from Brazil, with about 40 percent in India. Only about 2 percent was in the United States. India’s Internet users would soon adopt Facebook, leaving Orkut behind.
Mayer would later admit that if Orkut had been a bigger priority at Google, the company might have had more success with it in the United States and other countries. But in Google’s earlier days, “opportunity costs” determined how much attention a product received. In 2004, Google had 2,000 employees, around 800 of them engineers, dispersed in roughly a hundred teams of three to twelve each. Skimming twenty engineers each from a separate team meant that you would be losing an average of 15 percent of the manpower in those teams. As it was, twenty people had to be temporarily recruited that August to fix Orkut’s lingering problems. “I do think we made the right trade-off and the right balance,” Mayer would later argue. Considering how important social software would become, it was hard to agree.
Orkut was far from the only opportunity in the social sphere that Google missed. In May 2005, Google bought a small company in the promising area known as mobile social. Founded by Dennis Crowley and Alex Rainert, Dodgeball was a pioneering service that let mobile phone users turn their city into a giant hide-and-seek game where the
y could discover (or avoid) nearby friends. The possibilities of location-based services seemed endless, and tech followers applauded Google for its canny purchase.
But Google largely neglected its new prize. The tiny Dodgeball team was based in New York City, in order to keep in touch with the urban vibe of its product (Crowley and Rainert were veterans of New York University’s geek-hip Interactive Telecommunications Program). They constantly begged for attention and manpower from Mountain View, with little success. “It needed some love from Google to promote it,” Crowley later said. He felt strongly that if Google had paid even a little more attention, Dodgeball could have grown from 100,000 users to a million and more. At one point Sergey Brin visited the New York office and asked Crowley how things were going. “It’s awful,” Crowley told him. “We need more engineers.” Sergey said he’d get right on it. “But he didn’t,” Crowley later said. “Nothing came of it.”
Dodgeball had a feature called “shout” that let users broadcast a short message to their friends, and Crowley and his team thought of releasing a version of the program without locations—just those status updates. Stupid things like “I’m going to the movies.” One day in early 2006, a friend of Crowley’s from Nokia came in to Google and they shared their newest ideas. They slid their phones across the table and found they were working on the same idea—presence without location.
Not long afterward, Crowley saw the first version of an Internet start-up called Twitter. The company was run by Evan Williams, the cofounder of Blogger, who had sold that firm to Google in February 2003 but left in October 2004, unhappy at the relative neglect with which it treated his service. Twitter was a dead-simple Internet and phone service that let people broadcast 140-character messages to anyone who chose to “follow” the stray thoughts of a given user. Crowley began sending emails to people at Google telling them that this was important and Google should jump on it. “It all fell on deaf ears,” Crowley says. “They just weren’t interested in social at the time. It just wasn’t their thing.”
Crowley remembers a fateful videoconference with Mountain View in the summer of 2006 in which he and his colleagues argued that the social network movement was about to go crazy and now was the time to put more resources into Dodgeball. An executive flatly told him once and for all to forget about asking for more engineers. That sealed it for Crowley. Though reluctant to abandon Dodgeball’s loyal community, he and Rainert left in April 2007. It was two years before Google formally pulled the plug on the service. Meanwhile, Google would develop its own location-based service, Latitude. By then, there were a number of location-based start-ups, all of which owed something to Dodgeball. One of the hottest was called Foursquare.
Its cofounder was Dennis Crowley.
Google had a built-in disadvantage in the social networking sweepstakes. It was happy to gather information about the intricate web of personal and professional connections known as the “social graph” (a term favored by Facebook’s Mark Zuckerberg) and integrate that data as signals in its search engine. But the basic premise of social networking—that a personal recommendation from a friend was more valuable than all of human wisdom, as represented by Google Search—was viewed with horror at Google. Page and Brin had started Google on the premise that the algorithm would provide the only answer.
Yet there was evidence to the contrary. One day a Googler, Joe Kraus, was looking for an anniversary gift for his wife. He typed “Sixth Wedding Anniversary Gift Ideas” into Google, but beyond learning that the traditional gift involved either candy or iron, he didn’t see anything creative or inspired. So he decided to change his status message on Google Talk, a line of text seen by his contacts who used Gmail, to “Need ideas for sixth anniversary gift—candy ideas anyone?” Within a few hours, he got several amazing suggestions, including one from a colleague in Europe who pointed him to an artist and baker whose medium was cake and candy. (It turned out that Marissa Mayer was an investor in the company.) It was a sobering revelation for Kraus that sometimes your friends could trump algorithmic search.
In the summer of 2008, Kraus held a barbecue at his house for Googlers to kick around social networking ideas. He invited his long-term collaborator Graham Spencer. They also invited David Glazer, a forty-five-year-old Valley veteran who had recently been hired for the relatively senior role of engineering director.
Though the group discussed a number of ideas (“We could have renamed Orkut and painted it blue,” says Glazer), everyone agreed that Google should refrain from one path in particular: creating a “Facebook-killer” application of its own. “Google is terrible at being a taillight chaser,” Glazer later said. “Drunks follow taillights.”
Instead of building a Google product, they came to decide that the company should pursue a two-pronged strategy. One was making Google products more social—maybe Gmail and other applications could be opened up to people’s friends and contacts. The second was a more ambitious plan where Google would essentially create a scaffolding on the web to lubricate social activities. The system could duplicate some of the benefits of Facebook and Twitter without having people visit those websites. Kraus even had a motto for it: “go fast alone, go far together.” He also had a second slogan for the approach that Google had to take when competing in the social world: “Ready, fire, aim.” It sounded like a postmortem for a lost battle, he would later admit, but it was the Google way.
Google set about organizing many of the web’s socially oriented companies into a major initiative that it called OpenSocial. The idea was to build a shared infrastructure where multiple websites could participate in a more social web. A user’s identity would be portable; a profile formed on one site could be used on other sites or services. While Google bore the burden of much of the programming and organization, it was careful not to label the effort as solely its own: the party line was that this was an open-source group effort that would benefit all. But as some of the major participants—MySpace, Ning, hi5, Bebo, AOL—fell into line, the biggest social site sat out the effort. Facebook didn’t say it would never participate; it just didn’t. Meanwhile, a Facebook executive, Ethan Beard, emailed Joe Kraus to tell him about his company’s ban on sharing information with OpenSocial. Beard said that allowing movement of the personal information people shared with Facebook would be a violation of its terms of service agreement, even if a user wanted to share it.
Ultimately, Facebook’s lack of cooperation proved fatal to OpenSocial. You couldn’t duplicate Facebook without Facebook.
As OpenSocial lurched, Google began casting around for other ways to participate in social networking. One option was buying Twitter, but that was complicated by the fact that its founder, Evan Williams, had become disenchanted with Google by his previous experience as an immigrant through acquisition. Williams felt Google hadn’t developed Blogger to its fullest potential; though the blogging service had increased its audience, it had become lost among Google’s dozens of products and failed to innovate at its previous pace. In any case, Google was in its brief austerity period and was not in the mood to make a YouTube–level offer that Williams could not refuse. “This is not a time where I want to overpay,” said Schmidt in March 2009.
In theory, Twitter was so simple that Google could simply write its own version. “The question of the day was ‘Why don’t we build Twitter?’ Three guys could do it in a weekend!” said Glazer in 2009. But, he explained, that would have been a case of chasing taillights.
Google’s search team began working on improvements to its core technology that would allow for “social search” (based on signals of what your friends were searching for) and “real-time search” (which tried to respond to Twitter by raising the relevancy of fresh and popular sites—and indexing Twitter’s contents as quickly as people posted them). Another product was an ambitious communications service called Google Wave, created by a team working in isolation in the Google Australia office. At the May 2009 Google I/O conference its designers introduced the produc
t in a stunning ninety-minute demo that was posted on YouTube and became the talk of the web. But when the product began to appear in limited release later that year, it proved confusing to users. Wave required considerable instruction, but it was official Google policy not to provide that kind of support for its products. The demonstration offered by the team that developed Wave had wowed the crowd and impressed all who viewed it on YouTube thereafter. But how many users would take the time to sit through the ninety-minute demo before deciding to try the product? And even if they did, Wave would not be useful to them unless their friends and colleagues also knew how to surf the Wave. That was a lot to ask users to do on their own.
“I would say we were not executing well in the social space in general,” said Google VP Bradley Horowitz. “We had a bunch of different projects, but we didn’t have a coordinated goal that was going to get us in the conversation.”
In early 2009, Horowitz’s team began work on yet another new product that, Horowitz predicted, “would blow Twitter away.” Its code name was Taco Town, named after a Saturday Night Live parody of a Taco Bell commercial where a tortilla-covered snack is increasingly, and absurdly, slathered with more food. (“And it gets even awesomer when we take a deep-fried gordita shell, smear on a little of our special ‘guacamolito’ sauce, and wrap that around the outside!”) That reflected Googlers’ judgment of the Internet’s current social strategy: big, messy layers of greasy, unwholesome stuff whose caloric volume tried to compensate for satisfying essence. Taco Town was more focused. It was designed to work inside Gmail. (Giving it the advantage of instant exposure, though limiting it to only a plurality of netizens.) One of its definitive traits was the speedy process by which users would assemble their graph of friends—Taco Town would analyze email contacts and instantly present people with a social network that had been already built by their own behavior. Using that group as a starting point, Twitter-style comments (Tacos) could be posted—but, unlike in Twitter, the comments would not have a 140-character limit, and pictures and other media could be included in them. With a single mouse click within Gmail, you could replace the view of your inbox with a stream of Tacos from all your contacts.